diff --git a/egs/ami/ASR/pruned_transducer_stateless7/train.py b/egs/ami/ASR/pruned_transducer_stateless7/train.py
index b5efb3405..81823ced2 100755
--- a/egs/ami/ASR/pruned_transducer_stateless7/train.py
+++ b/egs/ami/ASR/pruned_transducer_stateless7/train.py
@@ -972,7 +972,16 @@ def run(rank, world_size, args):
         logging.info("Using DDP")
         model = DDP(model, device_ids=[rank], find_unused_parameters=True)
 
-    optimizer = ScaledAdam(model.parameters(), lr=params.base_lr, clipping_scale=2.0)
+    parameters_names = []
+    parameters_names.append(
+        [name_param_pair[0] for name_param_pair in model.named_parameters()]
+    )
+    optimizer = ScaledAdam(
+        model.parameters(),
+        lr=params.base_lr,
+        clipping_scale=2.0,
+        parameters_names=parameters_names,
+    )
 
     scheduler = Eden(optimizer, params.lr_batches, params.lr_epochs)
 
diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/train.py b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/train.py
index abfd56e5a..162ad8412 100755
--- a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/train.py
+++ b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/train.py
@@ -1036,7 +1036,16 @@ def run(rank, world_size, args):
         logging.info("Using DDP")
         model = DDP(model, device_ids=[rank], find_unused_parameters=True)
 
-    optimizer = ScaledAdam(model.parameters(), lr=params.base_lr, clipping_scale=2.0)
+    parameters_names = []
+    parameters_names.append(
+        [name_param_pair[0] for name_param_pair in model.named_parameters()]
+    )
+    optimizer = ScaledAdam(
+        model.parameters(),
+        lr=params.base_lr,
+        clipping_scale=2.0,
+        parameters_names=parameters_names,
+    )
 
     scheduler = Eden(optimizer, params.lr_batches, params.lr_epochs)